Welcome to our first online GAIA meetup!
As many of us are working from home and you can't meet in person, we wanted to fill the gap in social events and host an online meetup! This time we will meet people working in the Deep Learning teams at Zenuity.
There will be two talks during the evening, one about Active Learning and another about 3D Object Detection. For more information about the content, please see the abstracts below.
The meeting is hosted via Microsoft Teams and you can join from the desktop application, your browser or mobile device. You will be able to ask questions during the event through the Q&A chat session.
At Zenuity, we employ neural networks to solve a large number of perception tasks. Training these neural networks to a high accuracy requires a large amount of manually annotated data. Since annotation is quite costly, it is important that we do not waste money on annotating data that is not helpful for learning. ‘Active learning’ means that the decision of what to annotate next is guided by the neural network itself. In this talk, Erik will describe an experiment with active learning that we have performed at Zenuity, and show that this technique has a strong potential for the field of autonomous driving perception.
Detecting moving objects and determining their location relative to the car is a crucial part of enabling autonomous driving. In this talk, Adam will explain how we use camera and Lidar data to train models for detecting objects and predicting their location, orientation and size in 3D.
ZENUITY is a pure software company with the mission of elevating people’s lives by making self-driving cars real. Self-driving cars, when real, is estimated to reduce fatal car accidents with more than 85%, saving more than 50,000 persons from dying in car accidents caused by human errors each year. We feel that is a goal worth pursuing.
17:00 - Welcome to the online meetup with Zenuity!
17:05 - Active Learning for Autonomous Perception - Erik Werner
17:35 - Q&A
17:45 - 3D Object Detection - Adam Tonderski
18:15 - Q&A
18:30 - End of the event - thank you for joining
